<s>
Diagnostically	B-Algorithm
acceptable	I-Algorithm
irreversible	I-Algorithm
compression	I-Algorithm
(	O
DAIC	B-Algorithm
)	O
is	O
the	O
amount	O
of	O
lossy	B-Algorithm
compression	I-Algorithm
which	O
can	O
be	O
used	O
on	O
a	O
medical	B-Application
image	I-Application
to	O
produce	O
a	O
result	O
that	O
does	O
not	O
prevent	O
the	O
reader	O
from	O
using	O
the	O
image	O
to	O
make	O
a	O
medical	O
diagnosis	O
.	O
</s>
<s>
The	O
term	O
was	O
first	O
introduced	O
at	O
a	O
workshop	O
on	O
irreversible	B-Algorithm
compression	I-Algorithm
convened	O
by	O
the	O
European	O
Society	O
of	O
Radiology	O
(	O
ESR	O
)	O
in	O
Palma	O
de	O
Mallorca	O
October	O
13	O
,	O
2010	O
,	O
the	O
results	O
of	O
which	O
were	O
reported	O
in	O
a	O
subsequent	O
position	O
paper	O
.	O
</s>
<s>
The	O
"	O
amount	O
of	O
compression	O
"	O
in	O
irreversible	B-Algorithm
compression	I-Algorithm
used	O
to	O
be	O
determined	O
by	O
the	O
compression	O
ratio	O
,	O
where	O
the	O
acceptable	O
minimum	O
is	O
determined	O
by	O
the	O
algorithm	O
(	O
typically	O
JPEG	O
or	O
J2K	O
)	O
and	O
the	O
data	O
type	O
(	O
body	O
part	O
and	O
imaging	O
method	O
)	O
.	O
</s>
<s>
The	O
image	O
compression	O
community	O
has	O
long	O
used	O
objective	O
quality	O
metrics	O
like	O
SSIM	B-Algorithm
to	O
measure	O
the	O
effects	O
of	O
compression	O
.	O
</s>
<s>
In	O
the	O
absence	O
of	O
good	O
data	O
regarding	O
SSIM	B-Algorithm
,	O
the	O
ESR	O
review	O
of	O
2010	O
concluded	O
that	O
it	O
is	O
still	O
difficult	O
to	O
establish	O
a	O
criterion	O
for	O
whether	O
a	O
particular	O
irreversible	B-Algorithm
compression	I-Algorithm
scheme	O
applied	O
with	O
particular	O
parameters	O
to	O
a	O
particular	O
individual	O
image	O
,	O
or	O
category	O
of	O
images	O
,	O
avoids	O
the	O
introduction	O
of	O
some	O
quantifiable	O
risk	O
of	O
a	O
diagnostic	O
error	O
for	O
any	O
particular	O
diagnostic	O
task	O
.	O
</s>
<s>
An	O
2017	O
study	O
showed	O
that	O
a	O
SSIM	B-Algorithm
variant	O
called	O
4-G-r*	O
(	O
4-component	O
,	O
gradient	O
,	O
structural	O
component	O
of	O
SSIM	B-Algorithm
)	O
best	O
reflects	O
changes	O
in	O
images	O
that	O
affect	O
the	O
decision	O
of	O
radiologists	O
out	O
of	O
16	O
SSIM	B-Algorithm
variants	O
.	O
</s>
<s>
It	O
also	O
mentions	O
that	O
the	O
original	O
version	O
of	O
SSIM	B-Algorithm
works	O
as	O
poorly	O
as	O
a	O
basic	O
root-mean-square	O
distance	O
(	O
RMSD	O
)	O
for	O
this	O
purpose	O
,	O
a	O
result	O
echoed	O
by	O
the	O
2017	O
study	O
.	O
</s>
